Using Administrative Data in Health Economics: An Illustrative Study on Competition and Inequality Richard Cookson Centre for Health Economics University of York
Outline of talk 1. Background on administrative datasets used by health economists in England 2. Study of competition and inequality
Background on administrative health data in England 1. Patient level hospital data (HES) 2. Patient level primary care data (CPRD) Other useful administrative datasets I won t talk about include... QOF & GMS data on GP practices Clinical registry and audit data (e.g. cancer registry, national joint registry, central cardiac audit database etc.)
Hospital Episode Statistics All inpatient, outpatient and A&E attendances in NHS hospitals in England Inpatient data from 1989/90 Outpatient data from 2003/4 A&E attendances from 2007/8 Held by the Health and Social Care Information Centre on behalf of the Secretary of State Complex data extraction & manipulation to produce a useful dataset for research Stringent data protection requirements Data extraction fees www.hscic.gov.uk/hes
Patient information available Demographic Age, gender, ethnicity, and sensitive fields: NHS number, postcode Admission Dates, methods, source/destination, waiting time, length of stay Medical Diagnostic and procedure codes, treatment specialty, HRG for payment Death rates, emergency re-admission rates, patient reported outcomes Organisational Provider, commissioner, GP practice and sensitive field: consultant Geographical and socioeconomic Census output areas, wards, local authorities and attributed SES data Maternity Birth weight, gestation period, live birth, delivery methods Psychiatric Detention, mental health status, psychiatric status
Data generating process Hospital staff record information on the patient Medical record staff code this information and convert to electronic form Data sent to central warehouse Quality assurance Data linked to other sources of information Full data set compiled and made available
Centre for Health Economics 3-day Course on HES www.york.ac.uk/che/courses/short/patient-data
CLINICAL PRACTICE RESEARCH DATALINK (CPRD) www.cprd.com/intro.asp With thanks for his slides to my colleague Professor Tim Doran, University of York Department of Health Sciences
CPRD Clinical Practice Research Datalink ESTABLISHED 1987 OPERATED BY THE MEDICINES AND HEALTHCARE PRODUCTS REGULATORY AGENCY VISION SYSTEM PRACTICES ONLY (1,492-18.1%) CURRENTLY 545 ACTIVE PRACTICES (11.2M PATIENTS) MARKET SHARE OF THE MAJOR CLINICAL COMPUTING SYSTEMS VISION LV SYS ONE X PCS SYNERGY EAST MIDLANDS 6.7% 49.3% 27.7% 11.9% 3.5% EAST 22.4% 35.8% 2.2% 26.6% 6.0% LONDON 2.5% 24.8% 58.8% 12.0% 1.2% NORTH EAST 5.4% 35.4% 48.2% 6.5% 2.5% NORTH WEST 17.9% 42.0% 5.1% 22.9% 7.4% SOUTH CENTRAL 11.7% 33.9% 41.4% 6.1% 4.3% SOUTH EAST 26.0% 50.5% 1.6% 16.6% 2.5% SOUTH WEST 36.5% 36.8% 6.8% 10.9% 6.0% WEST MIDLANDS 27.2% 44.8% 4.4% 8.9% 6.9% YORKSHIRE HUMBER REGIONAL REPRESENTATION OF CLINICAL COMPUTING SYSTEMS 13.8% 42.6% 11.3% 8.0% 9.4%
BASIC CPRD DATA BASIC DATA EVENT FILES CLINICAL (ALL MEDICAL HISTORY DATA) REFERRALS IMMUNISATION THERAPY (INC. ALL PRESCRIPTIONS) TESTS (INC. RESULTS) LOOKUP FILES READ CODES (98,031 AVAILABLE) PRODUCT CODES (77,198 AVAILABLE) TEST CODES (304 AVAILABLE)
PATIENT LEVEL ANALYSES FOR DIABETES QUALITY OF CARE CPRD Example Diabetes Quality Trends ALLOWS ADJUSTMENT FOR AGE, SEX, CO-MORBIDITY, ETC (CF. PRACTICE LEVEL QMAS DATA)
Illustrative Study of Competition and Inequality
Three Doses of Hospital Competition in the English NHS Thatcher/Major Blair/Brown Cameron/Clegg 1991-7 2003-10 2010-?? 13
Project title: Funded by: Managed by: Effects of health reform on health care inequality NHS NIHR Service, Delivery and Organisation Programme DH PRP Health Reform Evaluation Programme Project duration: 1 April 2007-31 October 2010 Lead investigator: Richard Cookson Data analysis: Mauro Laudicella and Paulo Li Donni Advisory input: James Carpenter, Roy Carr-Hill, Diane Dawson, Mark Dusheiko, Hugh Gravelle, Geoffrey Hardman, Russell Mannion, Steven Martin, James Nelson-Smith, Andrew Street Special thanks: George Leckie and Carol Propper Department of Social Policy and Social Work The York Management School Department of Economics and Related Studies Yorkshire & Humber Public Health Observatory
Concerns that competition may undermine equity The availability of good medical care tends to vary inversely with the need for it in the population served. This inverse care law operates more completely where medical care is most exposed to market forces, and less so where such exposure is reduced. Dr Julian Tudor-Hart, 1971 (The Lancet) The commercialization of health care is the primrose path down which inexorably lies American medicine: first-rate treatment for the wealthy and 10th-rate treatment for the poor. Dr David Owen, 1989 (Quoted as leader of the opposition Social Democratic Party) Allowing private providers to compete for NHS business will exacerbate the inverse care law, because most profit can be made in more affluent healthier groups. Margaret Whitehead, Barbara Hanratty and Jennie Popay, 2010 (The Lancet)
Three stories 1. Competition undermines equity by reducing rent for benevolent hospitals to treat unprofitable patients 2. Competition undermines equity by crowding out benevolent hospital motivations 3. Competition improves equity by reducing waiting times and increasing activity in all hospitals, even poorly performing local hospitals disproportionately used by disadvantaged patients
Blair/Brown NHS Reforms 2001-8 Cookson, R, Laudicella, M and Li Donni, P. (2013). Does hospital competition harm equity? Evidence from the English National Health Service. 32(2): 410-422 Journal of Health Economics. http://dx.doi.org/10.1016/j.jhealeco.2012.11.009 Cookson, R, Laudicella, M, Li Donni, P and Dusheiko, M. (2012). Effects of the Blair/Brown NHS reforms on socioeconomic equity in health care. Journal of Health Services Research and Policy. 17(Suppl 1): 55 63; doi:10.1258/jhsrp.2011.011014
Blair/Brown NHS Reforms Sustained spending growth Real annual UK NHS expenditure growth averaged 6.56% from 1999/00 to 2010/11 compared with 3.48% from 1950/51 to 1999/00 Hospital reform Target driven performance management focusing especially on hospital waiting times Re-introduction of competition
Pro-competition elements of reform 1. Fixed price hospital payment (English HRGs) Piloted 2003/4 and fully implemented 2005/6 2. Patient choice of hospital Choice of 4-5 providers from December 2005 Free choice from 2008 3. Independent Sector (IS) entry ISTC programme share of overall NHS funded nonemergency activity grew from 0.02% in 2003/4 to 2.2% by 2008/9 (HES data) 11.94% for hip replacement, 5.29% for cataract Plus a substantial but unknown volume of sub-contracted IS activity
Market Concentration in England: 2003 2008 NHS Hospital Elective admissions Independent Sector Elective admissions HHI Index:
6,000 5,900 5,800 Did market concentration fall? Yes, a bit: -400 HHI pts (6.8%) 5,700 5,600 5,500 HHI NHS hospitals only HHI NHS and IS hosptals 5,400 5,300 5,200 2003 2004 2005 2006 2007 2008
Methods Basic regression design: difference-in-difference Compare the deprivation-utilisation gradient between more and less concentrated hospital markets, pre and post reform Time varying controls for population size, age-sex structure, disease prevalence, independent sector supply Improvement 1: Continuous treatment variable Avoids arbitrary split into groups Improvement 2: Year-by-year pattern of differences Expect gradual change as competition is phased in Improvement 3: Fixed effects Measure the dose of competition using change in market concentration, rather than the baseline level Improvement 4: Predicted market concentration index Predict market concentration using exogenous variables, to address potential endogeneity bias in models based on actual market concentration (2003 and 2008 only, as computationally intensive)
Non-emergency Inpatient Admissions By Dispersion and Deprivation 240 Non-deprived catch up in less competitive markets Affluent areas catching up in less competitive markets 220 200 180 160 140 High dispersion & deprived High dispersion & non-deprived Low dispersion & deprived Low dispersion & non-deprived Parallel Parallel growth growth in more in more competitive competitive markets markets 2003 2004 2005 2006 2007 2008 High dispersion refers to areas with HHI in 2003 < 5,000 (34.3% of areas) Deprived refers to areas with income deprivation > 20% (27.8% of areas)
Basic regression equation θ identifies the competition effect on inequality (deprivation * dispersion * post reform) y it is the utilisation count in small area i in year dispersionit is an index of market dispersion (=HHI * -1/100) deprivationi is the time invariant index of small area income deprivation I(t) is an indicator function of the post reform period xit is a vector of time varying controls, including need (small area population size and demographic characteristics and prevalence of diseases) and supply (number of independent sector hospitals within 60 km and whole time equivalent GP numbers) u i is the small area fixed effect
Level vs. change in hospital market concentration Level in 2003 Fall 2003-8
Main Finding No evidence that competition undermined socioeconomic equity in health care If anything, the opposite: deprived small areas experienced slightly faster growth relative to non-deprived small areas in dispersed (i.e. potentially more competition) markets However, this effect so small as to be economically unimportant
Back to the first two stories 1. Competition undermines equity by reducing rent for benevolent hospitals to treat unprofitable patients 2. Competition undermines equity by crowding out benevolent hospital motivations But are socially deprived hospital patients unprofitable under fixed price payment?
Hip replacement length of stay (allowing for other patient characteristics and hospital effects) 2001 2002 2003 2004 2005 2006 2007 (1) Most deprived decile 11.43 10.90 10.15 9.61 9.01 8.08 7.25 (2) Others 10.81 10.46 9.70 9.24 8.58 7.79 7.13 Gap: (1)-(2) 0.62 0.44 0.45 0.37 0.43 0.29 0.13 Ratio: (1)/(2) 1.06 1.04 1.05 1.04 1.05 1.04 1.02 (1) age 85 and over 16.62 15.62 14.87 14.88 13.82 12.45 11.81 (2) Others 10.56 10.25 9.50 9.01 8.33 7.55 6.89 Gap: (1)-(2) 6.06 5.37 5.37 5.87 5.48 4.89 4.92 Ratio: (1)/(2) 1.57 1.52 1.56 1.65 1.66 1.65 1.71 (1) 7 diagnoses or more 16.96 17.36 15.98 14.15 14.01 12.55 11.91 (2) Others 10.73 10.33 9.56 9.09 8.39 7.57 6.88 Gap: (1)-(2) 6.23 7.02 6.42 5.06 5.62 4.99 5.03 Ratio: (1)/(2) 1.58 1.68 1.67 1.56 1.67 1.66 1.73 Cookson, R, Laudicella M. Do the poor cost substantially more? The relationship between small area income deprivation and length of stay for elective hip replacement in the English NHS from 2001/2 to 2007/8. Social Science and Medicine 2011;72:173-84
Overall Conclusions Hospital competition in the English NHS in the 1990s and 2000s had little or no effect on socio-economic equity in health care Concerns about harmful equity effects proved to be exaggerated However, doses of competition were small Strong barriers to entry and exit Independent sector entry < 2.5% activity Public hospitals still tightly controlled
Thank you.